Oh boy, sparky, you've struck a nerve—but not an excruciatingly painful one. You ask an intuitive question. It’s SoapBox time
! Some electrical folks [across the board in experience] seem to have misconceptions about these systems, even though they have been in heavy use for a good part of the last century.
Warnings of their shortfalls have been published for at least half that time.
First—recognize the variations of "size" in ungrounded 480V systems. On the low end—15kVA bank serving an ancient 10hp hollow-shaft river-lift pump, a 5hp process-garbage screw conveyor, or a 2hp gear-drive fish ladder {via ~2 miles of tape-shielded EPR from the 13.8kV station bus in a 500kV switchyard.}. On the other extreme—an urban automobile- assembly plant with dozens of 3750kVA articulated secondary-unit substations; where each feeds several thousand feet of ‘sandwich’ bus duct.
Now—consider the correspondingly wide range of 480V-sysytem xero-sequence [“ground”] currents, based on the three naturally occurring capacitors that form from each phase to ground. Rule of thumb—the inescapable “charging” current that continuously flows from this characteristic is
ROUGHLY 300 milliamperes to maybe 3 amperes per 1000kVA of serving-transformer capacity.
The resistor should be able to effectively damp overvoltages on the 480-volt gear. Note that increased grounding-resistor current in excess of charging current is not normally of concern, but it allows unneeded additional resistor power dissipation with no other advantage.
Roger—there are some real gnarley looking toroidal bobbin-wound transformers made for that purpose.
